A Model Context Protocol server that provides Midjourney image generation capabilities through the GPTNB API.
Server defines 8 tools with complete JSON schemas and descriptions. Naming follows verb_noun convention (midjourney_imagine, midjourney_upscale, etc.). All tools have descriptions (194 - 250 chars, within baseline 34 - 392 range). Schemas include proper type definitions, enums, and constraints. However, parameter descriptions are generic and lack actionable guidance. Error handling is present but minimal, responses don't guide recovery or explain why failures occur. No tool annotations (readOnlyHint/destructiveHint) despite clear WRITE vs READ_ONLY risk classifications. Output schemas are not documented, forcing LLMs to infer response structure. Missing dependency hints (e.g., 'Call midjourney_get_task first to check status before upscaling'). Composition is sound, each tool has one responsibility, but chaining guidance is absent.
Execute a button action on a completed Midjourney task (like Zoom, Pan, etc.).
Blend 2-5 images together to create a new combined image.
Generate text descriptions of an image that could be used as prompts.
Get the status and results of a Midjourney task by ID.
Generate images using Midjourney with text prompts. Supports various parameters like aspect ratio, quality, and style settings.
Reroll/regenerate a Midjourney task with the same prompt but different results.
No tool annotations (readOnlyHint/destructiveHint/idempotentHint) despite clear WRITE vs READ_ONLY risk classifications in metadata. LLMs cannot distinguish safe reads from destructive writes without explicit hints.
Output schemas not documented. Handlers return createSuccessResult() with markdown text, but LLMs cannot parse task_id or status from unstructured responses. Downstream tools (midjourney_upscale, midjourney_get_task) require task_id but response structure is opaque.
Parameter descriptions lack actionable constraints and format guidance. E.g., 'seed' has no description of valid range or reproducibility semantics. 'action' parameter in midjourney_action accepts free-form strings with no enum or examples of valid button labels.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
Upscale a specific image from a completed Midjourney generation task.
Create variations of a specific image from a completed Midjourney generation task.
Error handling returns generic error messages without recovery guidance. E.g., 'Failed to submit blend task' does not tell LLM whether to retry, check prerequisites, or ask user for input. No error classification (retryable vs user-fixable vs fatal).
No dependency hints or chaining guidance. Tool descriptions do not explain prerequisites (e.g., 'Call midjourney_imagine first to generate images before upscaling') or suggest next steps. LLMs must infer workflow from names alone.